'Non-Partisan' comparison with vvvv and others, please!

Hi All,

I’m currently in the process of assessing the best platform for my team to focus on in the future and it’s currently between vvvv and Touch. I’ve been amazed that I can’t find any sensible articles or forum posts online about the strengths and weaknesses of the two. I’m also interested in the comparative capabilities of Open Frameworks, Max/MSP/Jitter, Cinder and others.

The purpose of this post is to ask genuine Touch users (some of whom I assume are fairly familiar with those other tools), to give as objective an opinion as possible. I really don’t want to turn this into a petty playground argument about the competing tools. I’m genuinely interested in understanding what the latest versions of these tools offer as we enter 2013.

Many thanks in advance for your combined wisdom!

Simon

P.S. I realise different tools suit different types of work so it might be useful to know where I’m coming from: We’re a production company specialising in interactive work. We do a lot of projects in visitor attractions, museums and with large corporates. Often this work involves communicating information, not just pretty images. Currently our core tools are Flash, .net, and, increasingly, Unity. We produce assets primarily in Premiere Pro, 3DS Max and After Effects. Sometimes we create large scale, multi-machine installations or shows. We’ve done this with custom developed Flash systems and, where projection mapping is required, we’ve used Catalyst (Mac).

No expert on vvvv, but from what i can tell is that currently, due to the larger user base, vvvv offers you more out of the box and plugin based features. But things are changing , and touch seems to catch up, especially if people start releasing more of their custom tools. This is where open frameworks and also vvvv have a big plus. People share a lot of their work.

All in all, I think touch is not as low level as vvvv, thereore the whole workflow is a little more artist friendly (while still giving you the possibility to create everything you can imagine) … It just feels more like you are working directly with the image.

A big plus for touch is the animation department and the possibility to create advanced GUIs for the final application and also for the internal tools you will create over the years.

Another one is the possibility for support of other platforms, as touch uses OpenGL, so one day we might see an OSX or Linux version, who knows. But it’s technically possible. with vvvv you will likely always be trapped in the windows world. But as vvvv uses direct x, this also means it works better with consumer level video cards.

A very big difference is support for video playback, which is so much better in touch.

So all in all I think touch is the perfect solution, especially if you have a mixed team of designers , animators and coders.

A ideal team could be
A technical director with perfect touch and python knowledge (creating custom tools and UIs)
A coder with c++ , python and glsl skills (for some advanced stuff, porting open frameworks classes to touch, …)
A bunch of designers, 3d artists, animators, all with basic to medium touch skills

Depending on how much time you invest into the pipeline setup you can actually create tools / workflows that allows untrained artists to create and finish interactive installations without much training in touch, something that I do not believe is possible in any of the other tools

Thanks for the reply Achim. I was particularly interested to read your points about graphics cards and video support. Food for thought.

Thanks again!

To elaborate a bit on the video performance equation, as Achim mentioned, TD is highly optimized for video playback. I think this goes a bit underappreciated as TD has a significant advantage - over 2x as efficient (in terms of raw CPU usage) vs. vvvv/jitter and over 3x as efficient vs. OpenFrameworks.

Wow. That’s both impressive and very specific! How did you come by those stats?

Is it the case that this kind of performance is reliant on a Gfx card costing $1000+ though?

Simon,

These findings were part of research conducted for a project where client asking for a performance comparison.

Results consistent across multiple codecs and tested on both Nvidia GeForce and AMD Radeon desktop cards. No need for the Pro models to achieve these figures.

As somebody whos dabbled in pretty much all of them (including the game engines too)

Personally I would go with touchdesigner unless you need a standalone app, whenever we’ve had to make a multiplatform app we’ve always gone with HTML5 and a bit of JQuery, you’d be amazed by how much you can achieve in the web app world these days, unity would pretty much be the equivalent of this I guess for your team.

With the setup above we’ve been able to manage everything, the old team I worked with was just myself (touch,games,web), a 3D/After Effects guy, a character/3D/drawing guy and another web/HTML5/Touch guy and I don’t think we ever really came across a project we technically couldn’t complete.

Thanks for all your feedback. In fact, precisely because of your feedback, I think it’s highly likely that we will be going with Touch as our main production environment from now on. Despite VVVV’s obvious community strengths, when I asked the same question on their forum I got zero responses. Not all that encouraging.*

My impression is that Touch is by far the most professional and professionally supported tool and that many of the more ‘open’ platforms almost kick back against an expectation that there are times a client demands answers and solutions.

I look forward to pestering you all for help in the future!

  • I have to revise this. It seems my post was in quarantine until my new username was authorised. The VVVV community actually came back with a fantastic bunch of responses too. It seems both products really are blessed with enthusiastic and generous communities.

I just saw you are about to setup a cluster, I think this is another interesting point for comparison between touch and vvvv.

Vvvv has a built in system which i hear works pretty well , and is quite easy to use. Don’t think it support perfect fame and buffer syncs, but it saves you from a lot of work

In touch you have to roll your own system. I really hope that derivaive will create a powerful cluster management workflow soon, but until then you have to build something that matches your needs

PS: There’s also gpu affinity in touch, a tech that allows you to bind touch to a specific quadro gpu, basically you create a cluster within a single host computer, with the advantage of better sync and easier file management…

And there is mosaic technology, an nvidia feature tat gives you up to 16 perfectly synced outputs per computer from a single touch instance. It’s not as fast as a cluster or affinity setup, but you save a lot of development time. Kinda depends on your project, it’s a very complex topic, especially if you want perfect frame sync …

Just to give you some idea of video performance in TouchDesigner on a beefy system. Recently benchmarking our HP Z800 with 12 cores and a Quadro K5000 I was easily able to stream 10 1080p videos encoded in H264 from 1 SSD. YMMV, but it is easy to surpass 3-4 HD videos with 1 SSD and a decent CPU setup. (multi-CPU cores are utilized for movie decoding in TouchDesigner)

These videos were in the mid-range bandwidth for H264.

TouchDesigner can handle tons of HD video. :smiley:

It sure can be done it touch, all I’m saying is that you gotta roll your own sync system and some admin stuff for reloading your scenes on the cluster clients.

Don’t worry about bandwidth on the nvidia solution.

Personally I used vvvv and MaxMSP/Jitter before TD (as well as some more imaging specific and non-realtime solutions such as Digital Fusion and Shake). I always found Max to be a bit annoying and insufficient, although certainly groundbreaking for when it was developed. vvvv has a very intuitive interface, but didn’t seem up to the task of handling intensive video tasks. TD is really unmatched as a video solution. Max is an audio program hacked to work with video, vvvv is more of a rendering/demo-scene type application. In this way I don’t know that there is really an apples to apples alternative.

I think the biggest issue with TD is that the interface still needs a bit of work (I wouldn’t call the workspace totally intuitive). It is easy to pick up for simple artistic tasks, but there is a bit of an uphill challenge when trying to go from beginner to more advanced capabilities. Also the rendernode/distribution licenses are somewhat unfriendly price-wise. The FTE version is really forward thinking though, and makes teaching/learning the software very accessible. vvvv is of course also available for free (or was last time I used it).

Bottom line is that TD is really in a class of its own for video. I do digital cinema work and had felt that certain things were not possible in an off-the-shelf realtime environment prior to TD. Stability wise, even though it is in Beta it is still more stable than some industry standard non-realtime compositors that have been around for ages. The developers are making very positive changes, 088 has really fantastic improvements over 077, so I would only expect that the gap would increase between TD and the competition.

I agree with all of the above, Derivative is a solid company that stands behind it users. It is production ready and there are many examples of it running solid 24x7 in major installations.

Here is my thread hijack, since we are discussing other software:
Watchout and the other systems that support multiple displays off of one system - how do they do with regard to sync and tearing? The watchout systems I have seen are pretty low powered systems with consumer video cards. Are the DirectX systems functioning better in this regard?

MD

Hello,

Watchout system is not really “realtime” as you have a production computer that will dispatch the project between the different display computers. You have a kind of rendering in this process. all computer will follow the production computer but you cannot make any change in realtime.

I think that media server like coolux , green hippo, catalyst have a working sync system within the network ( smpte, audioclock) that works pretty well

Xavier

WatchOut doesn’t render, no. Each server renders realtime, with the Production machine as the controller. Don’t know why WatchOut gets so much FUD from the media server community. If you want to work in a timeline and have a lot of screens, it’s a good choice.

There’s a few things that are changing with WatchOut sync. First is support for multiple outputs per computer (like ATI eyefinity cards), second is some tricks with audio clocking, I think.

But if your computers are fast enough, and you don’t have hard-butted screens, WatchOut doesn’t seem to have a problem.

Bruce